In order to perform the simulations of Asym2sig function deconvolution for BW
pyrolysis, fittings of Asym2sig function to Gaussian data (Fig.1(C)) were done firstly.
The non-linear fittings were performed according to Levenberg-Marquardt algorithm
via MATLAB® cftool tool. Figs.2 (A)-(C) gave the comparisons between calculated
Asym2sig values and Gaussian data of P-HC, P-CL and P-LG, respectively. These
figures demonstrated that the Asym2sig function could fit Gaussian data perfectly
with high R2 values. The values of parameters w2 equaled to w3, which was because of
the symmetrical feature of Gaussian profile. With regarded to profile width of w1, an
order of w1(P-LG)> w1(P-HC)> w1(P-CL) met the weight loss rate characteristics of
three pseudo components: DTG (-dm/dT vs. T profile) of P-HC was wider than P-CL
but narrower than P-LG; the width of P-LG profile was quite large (232.89K) and
stretched across the whole temperature range; w1 value of P-CL (w1=24.65K) was
more smaller than that of P-HC and P-LG. The narrow width of P-CL was attributed
to the crystal nature of cellulose (Hu et al., 2016). Once pyrolytic temperature reached
the peak temperature, cellulose decomposed fast and contributed to the abrupt slope
on right hand side of the whole DTG profile, as indicated in Fig.1(C).